How to Identify and draw an ellipse

[khoatran02]

Background

I have the image with the ellipse, and I am on a mission to calculate its area and draw it with precision.

Here's the original image:

The target image shows the desired result with the ellipse drawn:

Current Code

import cv2
import numpy as np
import matplotlib.pyplot as plt

# Apply CLAHE to the grayscale image
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
image_gray = cv2.imread("../data/results/phantom/capture_1_2024-07-09T15-42-41.png", cv2.IMREAD_GRAYSCALE)
image_gray = clahe.apply(image_gray)

# Threshold the image to create a binary image
_, binary = cv2.threshold(image_gray, 128, 255, cv2.THRESH_BINARY)

# Find contours in the binary image
contours, _ = cv2.findContours(binary, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)

# Find the largest contour by area
largest_contour = max(contours, key=cv2.contourArea)

# Fit an ellipse to the largest contour if it has enough points
if len(largest_contour) >= 5:
    ellipse = cv2.fitEllipse(largest_contour)
    (x, y), (major_axis, minor_axis), angle = ellipse

    # Check if the ellipse dimensions are valid
    if major_axis > 0 and minor_axis > 0:
        # Draw the ellipse on the image
        image_with_ellipse = cv2.cvtColor(image_gray, cv2.COLOR_GRAY2BGR)
        cv2.ellipse(image_with_ellipse, ellipse, (0, 255, 0), 2)

        # Calculate the area of the ellipse
        semi_major_axis = major_axis / 2
        semi_minor_axis = minor_axis / 2
        ellipse_area = np.pi * semi_major_axis * semi_minor_axis
        print(f"Ellipse parameters: Center=({x:.2f}, {y:.2f}), "
              f"Major Axis={major_axis:.2f}, Minor Axis={minor_axis:.2f}, Angle={angle:.2f}")
        print(f"Calculated Ellipse Area: {ellipse_area:.2f} square pixels")

        # Display the images
        plt.figure(figsize=(10, 5))
        plt.subplot(1, 3, 1)
        plt.imshow(image_gray, cmap='gray')
        plt.title('CLAHE Image')
        plt.subplot(1, 3, 2)
        plt.imshow(binary, cmap='gray')
        plt.title('Binary Image')
        plt.subplot(1, 3, 3)
        plt.imshow(cv2.cvtColor(image_with_ellipse, cv2.COLOR_BGR2RGB))
        plt.title('Result Image with Ellipse')
        plt.show()
else:
    print("No valid ellipse found in the largest contour")

Result figures

Problem

The ellipse in the image is not being properly identified, resulting in an incorrect calculation of the ellipse's area.

Question:

How can I correctly identify and draw the ellipse in the image?

[catunderrain]

Hi! I found that decreasing the threshold in the binary step to around 50 improves the results for your issue:

_, binary = cv2.threshold(image_gray, 50, 255, cv2.THRESH_BINARY)

The problem is that contours are continuous points, and with a high binary threshold, this continuity can break. Therefore, reducing the threshold helps maintain continuous zones in the pre-processed image.

I also try using morphology and increase contrast so that the zone seem more continuous with threshold 125:

clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
image_gray = clahe.apply(image_gray)

kernel = np.ones((5,5),np.uint8)

img3 = cv2.convertScaleAbs(image_gray, alpha=2, beta=0)
img32 = cv2.morphologyEx(img3,cv2.MORPH_OPEN, kernel)
img32 = cv2.morphologyEx(img32,cv2.MORPH_OPEN, kernel)
img32 = cv2.morphologyEx(img32,cv2.MORPH_OPEN, kernel)
img32 = cv2.morphologyEx(img32,cv2.MORPH_OPEN, kernel)

plt.subplot(1,3,1); plt.imshow(image_gray, cmap='gray')
plt.subplot(1,3,2); plt.imshow(img3, cmap='gray')
plt.subplot(1,3,3); plt.imshow(img32, cmap='gray')
plt.show()
image_gray = img32

I try with filter to decrease the thickness of zone so that we also see a better result with thes.125:

clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
image_gray = clahe.apply(image_gray)

kernel = np.ones((3,3),np.uint8)
kernel1 = np.array([[-0, -2, -0],
                    [-2,9,-2],
                    [-0,-2,-0]])

img3 = cv2.convertScaleAbs(image_gray, alpha=2, beta=0)
img32 = cv2.filter2D(src=img3, ddepth=-1, kernel=kernel1)
img32 = cv2.morphologyEx(img32,cv2.MORPH_OPEN, kernel)
# img32 = cv2.morphologyEx(img32,cv2.MORPH_OPEN, kernel)
# img32 = cv2.morphologyEx(img32,cv2.MORPH_OPEN, kernel)
# img32 = cv2.morphologyEx(img32,cv2.MORPH_OPEN, kernel)

plt.subplot(1,3,1); plt.imshow(image_gray, cmap='gray')
plt.subplot(1,3,2); plt.imshow(img3, cmap='gray')
plt.subplot(1,3,3); plt.imshow(img32, cmap='gray')
plt.show()
image_gray = img32

Anway, for better pre-processing of the image, you can apply filters such as edge detection, etc. to make contouring easier. GL bro!

[catunderrain]

Ok ông, là tui xóa hình trong reply của tui đúng k bro